* Do not use the BuildManager but have the qmakestep trigger
   the update via the target and project.

Reviewed-by: dt

 * Ease cross device development by introducing 'targets' which
   group build- and runsettings that are valid for this one target

 Most of the kudos for the code review go to dt. Con, thorbjorn,
 ckandler and others did also review parts of this patch.

Reviewed-by: dt

 * That is what everybody else is using now.

Reviewed-by: dt

 * This makes it easier to distinguish projects in the
   userfileaccessor

Reviewed-by: dt

QMake passes those headers directly to gcc, so the user doesn't need to
include themselves.

Reviewed-By: con

  * Use id() for methods returning a string used to represent
    some type of object.
  * Use displayName() for strings that are meant to be user
    visible.
  * Quieten some warnings while touching the files anyway.
  * Move Factories to their products in the plugins where that
    was not done before.

Reviewed-by: dt

This code was (more or less) duplicated in 4 places.

there is no point in throwing away and re-recreating invariant data over
and over ...

shaves off ~15% of the load time of Qt 4.6.

Should fix a lot of corner cases, might introduce a few bugs.
Also rename functions/slots to be better named.

Generic Project Manager and CMake Project Manager are missing from this
patch.

...making use of the new code model features. Move code to
separate file. Make it possible to obtain path to generated
header file from project manager.

Note: I didn't fix all the connects and there are a few missing things.
This compiles, more work is coming.

The pointers can be used to distinguish BuildConfigurations

More API work.

Each project has it's own BuildConfiguarion * classes, they'll get a
decent type safe interface and the setValue/value stuff will be removed.

Enables users to change which buildsteps get run per buildconfiguration.
Some further tweaks are probably necessary. This is a rather big
change, though it should work. :)

That is the effective qmake command and summary label.

Also ensure that Qt4ProjectConfigWidget does not emit any signals from
it's init method.

That is we actually parse the mkspec and evaluate QMAKE_CXX (and a few
other variables) to figure out the correct mkspec. This makes using
custom mkspecs easier and is also cleaner. I also changed mkspec() and
mkspecPath() to behave a little diffrent, essentialy mkspec() will
return only the name (the actual dir name) of the mkspec. That is in
general not sufficient for passing on to qmake. mkspecPath() only
returns the correct path to mkspecs/default.
Hopefully I haven't broken WinCE/Maemo/MinGW.

Reviewed-by: dt

It's easier this way for the unique naming.

Builds, but I'm pretty sure it doesn't *work* :)

Hmm even if the cached toolchain is correct, we do a few things, maybe
I should optimize it.

Made virtual newBuildConfiguration return a bool

This has still a few missing things, but this enough to start getting
some feedback. Missing are non qt projects, a solution for the
runconfiguration, a missing black line between the treeview on top and
the project settings at the bottom. Some flickering with removing/adding
widgets to the QScrollArea and not showing the expanded widget if the
Details button is right at the bottom.

Don't add -spec foo to the additional Arguments if foo is the same as
the default mkspec. This makes the common case better, while breaking if
you have QMAKESPEC set in teh environment and used -spec foo to override
it.

I need that in the Qt4ProjectConfigWidget to decide whether I should
show a import label.

We add each command line argument to the Additional Arguments

Create a new one and check if it's the same as the old one,
has the advantage that cached data doesn't need to be retrieved again.
It was that way before tool chain was moved to Qt4Project.